So I update to the latest 2.1.18. Mailman quits working. Sure, all the qrunners are running, but any mail to any list just disappears. Postfix gets it and sends it on:
May 11 19:09:51 autox postfix/local[22287]: 2A5C32D60BBD: to=<tn_admin@autox.team.net>, relay=local, delay=0.13, delays=0.01/0/0/0.13, dsn=2.0.0, status=sent (delivered to command: /local/mailman/teamnet/mail/mailman post tn_admin)
Go to the logs directory, grep for 'tn_admin' and nothing shows up. Where did the message go? Why is it not getting sent to the list? Why is there no trace of it in any logs?
mjb.
On 05/11/2014 06:20 PM, Mark J Bradakis wrote:
So I update to the latest 2.1.18. Mailman quits working. Sure, all the qrunners are running, but any mail to any list just disappears. Postfix gets it and sends it on:
May 11 19:09:51 autox postfix/local[22287]: 2A5C32D60BBD: to=<tn_admin@autox.team.net>, relay=local, delay=0.13, delays=0.01/0/0/0.13, dsn=2.0.0, status=sent (delivered to command: /local/mailman/teamnet/mail/mailman post tn_admin)
Go to the logs directory, grep for 'tn_admin' and nothing shows up. Where did the message go? Why is it not getting sent to the list? Why is there no trace of it in any logs?
What's in Mailman's error log?
Is incomingRunner running? OutgoingRunner?
Is the message archived?
What's in Mailman's qrunner log?
What's in Mailman's vette log?
What's in Mailman's queues (the various qfiles/* directories)?
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
Mark Sapiro wrote:
On 05/11/2014 06:20 PM, Mark J Bradakis wrote:
So I update to the latest 2.1.18. Mailman quits working. Sure, all the qrunners are running, but any mail to any list just disappears. Postfix gets it and sends it on:
May 11 19:09:51 autox postfix/local[22287]: 2A5C32D60BBD: to=<tn_admin@autox.team.net>, relay=local, delay=0.13, delays=0.01/0/0/0.13, dsn=2.0.0, status=sent (delivered to command: /local/mailman/teamnet/mail/mailman post tn_admin)
Go to the logs directory, grep for 'tn_admin' and nothing shows up. Where did the message go? Why is it not getting sent to the list? Why is there no trace of it in any logs?
What's in Mailman's error log?
Nothing.
Is incomingRunner running? OutgoingRunner?
Like I said, the runners are running.
Is the message archived?
No.
What's in Mailman's qrunner log?
Nothing.
What's in Mailman's vette log?
Nothing.
What's in Mailman's queues (the various qfiles/* directories)?
The message was in qfiles/in and once I reverted back to 2.1.14, rebooted and restarted mailman it got delivered.
On a possibly related note, when I ran mailmanctl stop the 2.1.18 IncomingRunner did not die. Doing a kill -KILL on it would kill it, but it would immediately respawn. After several attempts to get rid of it, I installed 2.1.14 and rebooted the server.
mjb.
On 05/12/2014 01:40 AM, Mark J Bradakis wrote:
Mark Sapiro wrote:
Is incomingRunner running? OutgoingRunner?
Like I said, the runners are running.
I apologize for my skepticism, but many people will see a list of qrunner processes and assume that *all* are running.
What's in Mailman's qrunner log?
Nothing.
There can't be "Nothing" in the qrunner log unless you never started mailmnanctl or you're looking at the wrong log?
The message was in qfiles/in and once I reverted back to 2.1.14, rebooted and restarted mailman it got delivered.
Then IncomingRunner wasn't processing the queue. It may have died, or due to some install glitch it may have been looking at a different in/ queue from the one where the process invoked by Postfix put it. Maybe the install of 2.1.18 put things in different places so the Postfix aliases pointed to the 2.1.14 install and not 2.1.18. That might also explain why the qrunner log was empty if it truly was.
On a possibly related note, when I ran mailmanctl stop the 2.1.18 IncomingRunner did not die. Doing a kill -KILL on it would kill it, but it would immediately respawn. After several attempts to get rid of it, I installed 2.1.14 and rebooted the server.
mailmanctl stop depends on Mailman's data/master-qrunner.pid containing the PID of mailmanctl. Perhaps things were confused about where this file was.
The respawning is probably due to some watcher that's part of some previously installed package.
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
Mark Sapiro wrote:
What's in Mailman's qrunner log?
Nothing.
There can't be "Nothing" in the qrunner log unless you never started mailmnanctl or you're looking at the wrong log?
My turn to apologize, I should have specified that the logs contained nothing related to that specific message, and not implied they were completely empty.
The message was in qfiles/in and once I reverted back to 2.1.14, rebooted and restarted mailman it got delivered.
Then IncomingRunner wasn't processing the queue. It may have died, or due to some install glitch it may have been looking at a different in/ queue from the one where the process invoked by Postfix put it. Maybe the install of 2.1.18 put things in different places so the Postfix aliases pointed to the 2.1.14 install and not 2.1.18. That might also explain why the qrunner log was empty if it truly was.
I'll try reinstalling the latest and greatest mailman, see if I can figure out what IncomingRunner is actually doing.
mjb.
PS: This is on a server running Ubuntu 13.10
On 05/12/2014 05:51 PM, Mark J Bradakis wrote:
My turn to apologize, I should have specified that the logs contained nothing related to that specific message, and not implied they were completely empty.
Please note that when I ask what's in logs, I don't want a filtered response. Granted there's a bunch of stuff in logs that isn't relevant, but I would like it not pre-filtered. The bottom line is if you knew what would be relevant to me, you probably wouldn't need my help.
For example, you indicated in the OP that you grepped for the list name. There could be relevant messages that don't include the list name
-- Mark Sapiro <mark@msapiro.net> The highway is for gamblers, San Francisco Bay Area, California better use your sense - B. Dylan
Mark J Bradakis writes:
So I update to the latest 2.1.18.
"2.1.18" is not the latest and is known to have annoying bugs (this may be one of them). Did you mean "2.1.18-1"?
Aside to Mark: I think even for a brown-bag release it would be best to bump the micro version (in this case to "19" instead of "18-1").
Go to the logs directory, grep for 'tn_admin' and nothing shows up. Where did the message go?
It's probably in a queue, most likely "shunt" since it's not being delivered, but it could be somewhere else if one of the qrunners is broken.
Sorry I can't be more precise but wanted to provide what help I can quickly.
Steve
participants (3)
-
Mark J Bradakis -
Mark Sapiro -
Stephen J. Turnbull